#####################################
These files are part of the dataset
"Historical building footprint area (BUFA) - gridded surfaces for the U.S. from 1900 to 2010",
contained in the dataverse "HISDAC-US: Historical Settlement Data Compilation for the United States" 
(https://dataverse.harvard.edu/dataverse/hisdacus).

#####################################
Reference:
Uhl, Johannes H.; Leyk, Stefan, 2022, "Historical building footprint area (BUFA) 
- gridded surfaces for the conterminous U.S. from 1900 to 2010", 
https://doi.org/10.7910/DVN/HXQWNJ, Harvard Dataverse,

#####################################
Description:
Each GeoTIFF file contains a geospatial raster layer with pixel values representing the total building footprint area 
i.e., the built-up surface in m²) within each 250 m x 250 m grid cell, at 10 years temporal resolution. 
Data sources: Microsoft USBuildingFootprints v2019 + Zillow Transaction and Assessment Dataset (ZTRAX, Zillow Inc., v2017). 
USBuildingFOotprints were obtained from https://github.com/microsoft/USBuildingFootprints; ZTRAX data was obtained via data share agreement between Zillow, Inc., and University of Colorado Boulder.

#####################################
Methodology:
- We created point-based geospatial vector data from ZTRAX property records, attributed with the construction year of each built-up structure (YearBuilt).
- For each county, we obtained the building footprint data from the USBuildingFootprints dataset.
- We spatially joined the ZTRAX point features and building footprint polygons, by assigning the nearest ZTRAX record to each building footprint polygon.
- This way, we transferred the YearBuilt attribute from ZTRAX to the building that most likely represents the building described in ZTRAX.
- We then stratified the YearBuilt-attributed building footprint data in decades, and calculated the total area of the building footprint polygons falling within each grid cell of the HISDAC-US grid, per decade.
- Finally, we accumulated the total building footprint area newly built-up per decade, to get gridded surfaces measuring the total built-up area per grid cell and year.
Related code:
https://github.com/johannesuhl/ztrax2sqlite2csv
https://github.com/johannesuhl/geoprocessing

Difference between the HISDAC-US BUA layers and BUFA layers:
The BUA layers ("built-up areas") are a set of binary surfaces, with grid cells of value "1" indicating a 250x250m cell that is "developed", i.e.,
that contains at least one built-up property from ZTRAX in a given year, and "0" undeveloped grid cells.
The BUFA layers ("building footprint area") reports for each grid cell the total area within the grid cell that is actually built-up (i.e., the sum of the building footprint area per grid cell in sqm). 

Limitations and uncertainties:
There may be discrepancies between BUA and BUFA due to the different vector data used for rasterization (i.e., ZTRAX locations for BUA, and Microsoft building footprints for BUFA).
Moreover, the spatial join rule used may produce incorrect matches, i.e., attributing a building footprint with an incorrect YearBuilt.
Furthermore, buildings that fall within regions not covered by ZTRAX or without YearBuilt coverage, are omitted in the resulting BUFA surfaces.
Lastly, these layers represent estimates of historical building footprint area per grid cell, without taking into account historical changes in building footprint area, nor taking into account buildings that were replaced or torn down.
In other words, the contemporary building footprint geometries combined with the construction years of the contemporary building stock are used to model historical building footprint area per grid cell.
In areas of heavy urban renewal building replacement, or urban shrinkage, the provided estimates may be biased.

To quantify these effects, users are refered to the following data / references:

Uhl, Johannes H.; Leyk, Stefan, 2020, "Uncertainty surfaces accompanying the BUPR, BUPL, and BUA gridded surface series", https://doi.org/10.7910/DVN/T8H5KF, Harvard Dataverse, V1 

Uhl, J. H., Leyk, S., McShane, C. M., Braswell, A. E., Connor, D. S., & Balk, D. (2021). Fine-grained, spatiotemporal datasets measuring 200 years of land development in the United States. Earth system science data, 13(1), 119-153.

#####################################
Data sources: Zillow Transaction and Assessment Dataset (ZTRAX) (c) Zillow Inc. & Microsoft USBuildingFootprints v2019.
Spatial resolution: 250m
Coordinate reference system:
USA Contiguous Albers Equal Area Conic USGS version (SR-ORG:7480)
https://spatialreference.org/ref/sr-org/usa_contiguous_albers_equal_area_conic_usgs_version-2/
Proj4: +proj=aea +lat_1=29.5 +lat_2=45.5 +lat_0=23.0 +lon_0=-96 +x_0=0 +y_0=0 +ellps=GRS80 +datum=NAD83 +units=m +no_defs
Note that SR-ORG:7480 is identical to ESRI:102039, SR-ORG:8976, and EPSG:5070.

Contact:
Stefan Leyk
Department of Geography
University of Colorado Boulder
GUGG 110, 260 UCB 
Boulder, CO 80309-0260, United States of America
stefan.leyk@colorado.edu
